Hi John,

> The attached patch fixes it for me. The cause seems to be that the
> returned alist for a directory browser has entries for (directory
> ...) and (Last-Modified ...) and the code was only getting rid of
> the former only. That leftover (Last-Modified ...) would then trip
> up the assert in mpc--proc-alist-to-alists when it checks with
> mpc--proc-alist-to-alists-starters.
[...]
> diff --git a/lisp/mpc.el b/lisp/mpc.el
> index 4317fece6fc..05ef546f884 100644
> --- a/lisp/mpc.el
> +++ b/lisp/mpc.el
> @@ -644,9 +644,10 @@ mpc-cmd-find
>                          (mpc-proc-buf-to-alist
>                           (mpc-proc-cmd (list "listallinfo" value)))))
>                     (mpc--proc-alist-to-alists
> -                    ;; Strip away the `directory' entries.
> +                    ;; Strip away the `directory' & `Last-Modified' entries.
>                      (delq nil (mapcar (lambda (pair)
> -                                        (if (eq (car pair) 'directory)
> +                                        (if (or (eq (car pair) 'directory)
> +                                                (eq (car pair) 'Last-Modified))
>                                              nil pair))
>                                        pairs)))))
>                  ((string-match "|" (symbol-name tag))

Hmm... IIUC the output we receive is a list of pairs made of sublists of
the form either:

    (directory . <...>)
    (Last-Modified . <...>)

or

    (file . <...>)
    (Last-Modified . <...>)
    (Format . <...>)
    (Title . <...>)
    ...

IIUC when I wrote the code there were no such `Last-Modified` entries.

We're interested in extracting a list of the files (where each file is
represented by its alist), so I just stripped away all the `directory`
entries and that was it.  But now that `directory` can be followed by
info about that directory (currently only `Last-Modified`), we should
arguably be more careful to also remove everything between
`directory` and the next `file` or `directory`.

So I think the patch below is "more robust".
WDYT?
I also think it's a good opportunity to add some comments/docstrings
because it took me a while to understand what this was trying to do.


        Stefan